Ramp launches Router.com to optimize AI inference costs

Ramp, a corporate spend platform, has launched Router.com, a single API endpoint designed to optimize AI inference costs. The tool functions by matching each AI request to the lowest-cost model that satisfies a developer’s specific performance and quality requirements.

According to the company, existing customers have seen an average reduction in inference costs of 40%. The service currently supports 27 models, including offerings from OpenAI, Anthropic, and SpaceXAI, with support for Gemini expected soon. Open-weight models from providers like Nvidia, Kimi, DeepSeek, GLM, and Qwen are also available through various providers.

Rahul Sengottuvelu, Ramp’s chief technology officer, noted that AI is one of the fastest-growing and least measurable expenses for most companies. Ramp is offering free routing through 2026, with users paying only for the tokens they consume. The company reported that its own internal use of this technology has already reduced inference costs by approximately 30% while maintaining over 99.9% reliability.